Top: John S. Allen's Home Page
Up: Table of Contents
Previous: Response by John S. Allen
Next: Calculation of proportion of bicycles


[John Forester is the founder of the Effective Cycling program (the first scientifically based program of instruction in safe bicycling), author of the groundreaking books Effective Cycling and Bicycle Transportation, and serves as an expert witness in bicycle accident lawsuits. He has posted extensive comments about the CPSC's bicycle regulations on on his Web site. -- JSA]

John Forester's comments (1996)

I agree with much of John Allen’s reply to Rodgers, but I think there’s more to say.

1.264 mph is not a credible average speed. Rodgers suggests that the median speed is more credible as a measure of central tendency, advancing a grand total of 20 miles per month achieved in 9 hours, a median speed of 2.2 mph. That is equally incredible.

The Rodale and the CPSC data do not agree, as the average speed calculation  shows. But which set of data is more accurate? Rodgers prefers the CPSC data, effectively dismissing any comparison with the Rodale mileage data and data from all other studies, including Kaplan’s, which examined a population that paid attention to distance ridden, and which was supported by odometer measurements.

Merely comparing any single study against the CPSC study can not determine which is more accurate. That choice depends on other comparisons between the data sets and the real world. The CPSC data show up very badly, because the other studies have large areas of agreement, and because the CPSC data paint pictures that are obviously inaccurate, such as the proportion of bicycles among highway traffic (see calculations on next page) and the risk ratios between types of facility.

Allen criticized the CPSC for not seeking the advice of "experts on bicycling research." Rodgers limited his reply, saying only that the choice of hours instead of miles as the measure of exposure was made after consultation with "bicycle experts." Rodgers then named these: the Bicycle Federation of America, the Bicycle Manufacturers Association, the National Highway Traffic Safety Administration, the Federal Highway Administration, and the National Safe Kids Coalition. That statement demonstrates the paucity and inaccuracy of the CPSC's knowledge: bicycle experts are not bicycling experts, and none of these organizations is known for research on bicycling.

Much of the debate concerns the choice of measure of exposure, hours or miles. Rodgers argues that hours can be measured more accurately than miles, given that most people know the time of day while few people have odometers on their bicycles But were the hours counted accurately? Given that the CPSC data disagree with other obvious data, it appears not.

We also should ask for what kinds of policy hours of use is a better measure than distance traveled. If bicycling is merely a recreational pastime, then any changes that reduced the accident rate per hour would be beneficial. However, even many recreational cyclists seek distance traveled, not time awheel, whether their aim is touring or physical exercise. Other recreational cyclists consider the time spent riding to their chosen recreational sites as waste rather than as benefit.

The obvious implementation of the CPSC's study for transportation is to build bike paths for slow-speed use. Although Rodgers denies this conclusion in his reply to Allen, the CPSC study specifically recommends that to increase safety "bike paths [defined as sidepaths, the most dangerous kind of all] and bike lanes should also be considered," and, "Environmental factors might be addressed by improving road design, or by promoting the development of bike lanes and bike paths." If those words are not recommendations for lanes and paths, then what are they? They certainly do not suggest that we need further studies to determine whether these facilities are much safer than good roads.

Now, to illustrate the mathematics, suppose that the average bike path route requires twice the trip time as the road route it replaces, and produces an equal number of accidents. The amount of transportation and the number of accidents are the same, but the accident rate, per hour of exposure is halved. The comparison only holds if the same number of people cycle. It is highly unlikely that present transportational cyclists would choose to halve their speed and double their travel time; many would drive instead. Undoubtedly, time awheel is the worst measure of cycling for transportational purposes that could be imagined.

Rodgers mentions the Federal aim of doubling the amount of cycling transportation. He does not recognize what should have been obvious, that a policy based on data relevant to recreational cycling is not appropriate for transportational cycling. Rodgers states that the typical American cyclist does not behave like the typical member of the League of American Bicyclists. Well, it would be far better if those recruited to cycling transportation learned to behave like L.A.B. members, who have the lowest accident rate that has been measured. One way to achieve this result is to have a policy based on transportational cycling instead of recreational cycling.

There is a clear incompatibility between different parts of the CPSC data on risk ratios between different facilities, as well as between the CPSC data and that of other studies. The CPSC data show a far higher cycling accident rate on roads than on paths, even though car-bike collisions are only a small fraction of accidents to cyclists. Then, either most accidents on roads are caused by factors other than motor vehicles that appear on roads but not on paths, or the data are incorrect. Rodgers does not merely ignore this incompatibility,  he defends recommendations based on ignoring it.

Generally, paths are rated as more dangerous than roads. There is not one non-motor-vehicle physical factor in which roads have been identified as more dangerous than paths. There is even a behavioral factor in which paths are rated as more dangerous than roads: the chaotic nature of bike-path traffic makes cycling at normal road speeds extremely dangerous.

Given this contradiction, it is entirely wrong to recommend building bike paths because of their physical attributes, because these obviously do not produce the relatively low accident rates per hour of use.

This leaves the possibility that the difference is behavioral, as I stated in my work of twenty years ago. The safe speed on paths is very slow, but most cyclists on paths ride within the envelope of safe speed. The exceptions are the cyclists who normally ride fast. The Kaplan data indicate that, for a population of American cyclists with a higher average speed (and lots more skill) than the complete American cycling population, bike paths are 2.6 times as dangerous as average roadways. The CPSC data apply only to present, largely recreational, cyclists. They do not demonstrate that cyclists transferred from road to path will adopt the same low speed, or that cyclists who transfer from the recreational mode to the transportational mode will continue to ride slowly. For these reasons too, hours of use is the wrong measure of cycling, and practices and recommendations based on the CPSC study are inappropriate for transportation policy.

Relevance to the CPSC’s mission

The prime purpose of the CPSC, as established by Congress, is reducing casualties caused by defective product design. The mandate related to bicycling can be conveniently divided into nighttime protective equipment and everything else.

Rodgers states that the nighttime fatalities disclosed by the study stimulated the CPSC to hold its 1994 meeting on that subject. The inadequacy of the CPSC's requirements for nighttime protective equipment had been conclusively demonstrated twenty-five years ago, yet it took a body count to raise any interest on the part of the CPSC in revising its requirements.

The study concluded that the existing CPSC standard adequate in all other areas because there were few product failure type accidents, approximately the same low number as before the CPSC requirements were issued. This attitude reverses the attitude of the CPSC in past years.

When I sued the CPSC over its regulation, back in the 1970s, I stated that many of the requirements were unlawful because they addressed conditions that did not cause a significant number of injuries, and did not fix whatever might be the problem. I also stated that the requirements, specifying the type of chainguard on children's bikes and the all-reflector system for nighttime protection, addressed known causes of injuries yet were unlawful because they adopted inadequate cheap fixes in place of the adequate countermeasures that had been used in many parts of the world for decades.

The CPSC replied twenty years ago that its scientists had expert knowledge of accidents and their causes and did not need a body count before issuing regulations; indeed, that they should not let people die while waiting for casualties to occur. Nowadays, in contrast, the CPSC uses body counts to defend the parts of its regulation that never addressed the causes of accidents, and also to alert itself to nighttime accidents. Incidentally, a study by Russ Petty of the Babson Institute showed that the accident rate per bicycle has risen as the proportion of bicycles conforming to the CPSC requirements has risen; not a recommendation for the effectiveness of the regulation.

Children’s ability to learn safe cycling behavior

Neither Allen nor Rodgers discussed the role of cyclist training in reducing accidents. (Allen remarked only that the CPSC study downplayed training.) The CPSC study, however, stated that studies of child development showed that children below the age of 10 do not have the maturity to learn safe cycling behavior; they dart out of driveways because they cannot understand that motorists might also want to use the same road. The study then said that "most children by the sixth grade have the ability to understand and perform the taught behaviors." Despite more than a page of references, the CPSC cited not one study of actual attempts to train children in proper cycling practices.

My work with child cyclists showed that eight-year-olds can be trained to follow the basic traffic cycling principles on two-lane residential streets; ten-year-olds, on multi-lane streets. The work was published by the Transportation Research Board and is well known to cycling experts; the CPSC should have been able to find it.

The ability to ride safely can be developed in much younger children. Just yesterday I was riding with my five-year-old granddaughter on her first outing on normal city streets. She had just learned to stay upright while riding on playground surfaces. In less than one hour's riding around our neighborhood she had learned which was the proper side of the road (although she was still vague about the words right and left), to ask me when approaching an intersection whether any cars were coming, and to stop at stop lines, to walk her bicycle forward to the position from which I could observe traffic on the major road, and to restart when I said that no cars were coming, an observation that I had her confirm. This is not abstruse knowledge; as I remarked twenty years ago, typical club cyclists can cite many examples of seven-year-olds riding well with their parents, even going on club rides. In other words, training child cyclists is both much easier and can be done at much earlier ages than the CPSC indicates, and this information has been readily available for decades.

Next


Top: John S. Allen's Home Page
Up: Table of Contents
Previous: Response by John S. Allen
Next: Calculation of proportion of bicycles

Contents 1996, John Forester
Last modified April 19, 2001